Non-Binary values taken by binary variables in GAMS/LINDO

Solver related questions
Post Reply
sidsharma
User
User
Posts: 5
Joined: 6 years ago

Non-Binary values taken by binary variables in GAMS/LINDO

Post by sidsharma »

Hi experts,

I am new to GAMS and modelling a stochastic optimization problem using GAMS/LINDO.
In the output returned by the solver, the variables I declared as binary are taking real values between 0 and 1 (eg. 0.66).
Can someone please explain to me the scenario in the context of GAMS/LINDO.
Following is the model stats and solver summary.

MODEL STATISTICS

BLOCKS OF EQUATIONS 7 SINGLE EQUATIONS 131
BLOCKS OF VARIABLES 6 SINGLE VARIABLES 18,951
NON ZERO ELEMENTS 26,231 NON LINEAR N-Z 12,000
DERIVATIVE POOL 20 CONSTANT POOL 16
CODE LENGTH 18,030 DISCRETE VARIABLES 18,750


GENERATION TIME = 0.047 SECONDS 6 MB 25.1.1 r66732 WEX-WEI




S O L V E S U M M A R Y

MODEL NFV OBJECTIVE total_sc_count
TYPE EMP DIRECTION MINIMIZE
SOLVER LINDO FROM LINE 152

**** SOLVER STATUS 1 Normal Completion
**** MODEL STATUS 1 Optimal
**** OBJECTIVE VALUE 20.5000

RESOURCE USAGE, LIMIT 7.343 1000.000
ITERATION COUNT, LIMIT 15290 2000000000
EVALUATION ERRORS NA 0

The value of the binary variable taken by LINDO:

LOWER LEVEL UPPER MARGINAL

1 .p1 . 0.5000 1.0000 EPS
1 .p2 . 0.3333 1.0000 EPS
1 .p3 . 0.5000 1.0000 EPS
2 .p1 . 0.8333 1.0000 EPS
2 .p2 . 0.8333 1.0000 EPS
2 .p3 . 0.8333 1.0000 EPS
3 .p1 . 0.6667 1.0000 EPS
3 .p2 . 0.6667 1.0000 EPS
3 .p3 . 1.0000 1.0000 EPS
4 .p1 . 0.6667 1.0000 EPS
4 .p2 . 0.6667 1.0000 EPS
4 .p3 . 0.6667 1.0000 EPS
5 .p1 . 0.8333 1.0000 EPS
5 .p2 . 0.8333 1.0000 EPS
5 .p3 . 0.8333 1.0000 EPS
6 .p1 . 0.8333 1.0000 EPS
6 .p2 . 0.8333 1.0000 EPS

Thanks and regards,
Sidharth
User avatar
Renger
Posts: 639
Joined: 7 years ago

Re: Non-Binary values taken by binary variables in GAMS/LINDO

Post by Renger »

Hi
Perhaps this might help you (same issue when using dicopt: viewtopic.php?f=13&t=6802&p=15100&hilit ... ous#p15100.
Cheers
Renger
____________________________________
Enjoy modeling even more: Read my blog on modeling at The lazy economist
Post Reply